| 1 |
# (C) Copyright 2005-2023 Enthought, Inc., Austin, TX |
| 2 |
# All rights reserved. |
| 3 |
# |
| 4 |
# This software is provided without warranty under the terms of the BSD |
| 5 |
# license included in LICENSE.txt and may be redistributed only under |
| 6 |
# the conditions described in the aforementioned license. The license |
| 7 |
# is also available online at http://www.enthought.com/licenses/BSD.txt |
| 8 |
# |
| 9 |
# Thanks for using Enthought open source! |
| 10 |
|
| 11 |
""" |
| 12 |
Property Traits |
| 13 |
=============== |
| 14 |
|
| 15 |
The ``Image`` class has three traits which are closely related: ``scan_size``, |
| 16 |
``scan_width`` and ``scan_height``. We would ideally like to keep all of these |
| 17 |
synchronized. This can be done with trait observation, as shown in the |
| 18 |
previous section, but this sort of pattern is common enough that Traits has |
| 19 |
some in-built helpers. |
| 20 |
|
| 21 |
Instead of declaring that ``scan_width = Float()`` we could instead declare it |
| 22 |
to be a ``Property`` trait type. Property traits are similar to ``@property`` |
| 23 |
decorators in standard Python in that rather than storing a value, they compute |
| 24 |
a derived value via a "getter", and optionally store values via a "setter". |
| 25 |
Traits uses specially named methods of the form ``_get_<property>`` and |
| 26 |
``_set_property`` for these "getters" and "setters." If there is a "getter" |
| 27 |
but no "setter" then the property is read-only. |
| 28 |
|
| 29 |
Additionally, we need to know when the value of the property might change, and |
| 30 |
so we need to declare what traits to observe to know when the property might |
| 31 |
change. What all this means is that we can define ``scan_width`` as a |
| 32 |
property by:: |
| 33 |
|
| 34 |
class Image(HasTraits): |
| 35 |
... |
| 36 |
|
| 37 |
scan_width = Property(Float, depends_on='scan_size') |
| 38 |
|
| 39 |
def _get_scan_width(self): |
| 40 |
return self.scan_size[0] |
| 41 |
|
| 42 |
def _set_scan_width(self, value): |
| 43 |
self.scan_size = (value, self.scan_height) |
| 44 |
|
| 45 |
Traits will then take care of hooking up all the required observers to make |
| 46 |
everything work as expected; and the `Property` can also be observed if |
| 47 |
desired. |
| 48 |
|
| 49 |
Simple `Property` traits like this are computed "lazily": the value is only |
| 50 |
calculated when you ask for it. |
| 51 |
|
| 52 |
Cached Properties |
| 53 |
----------------- |
| 54 |
|
| 55 |
It would be quite easy to turn the histogram function into a read-only |
| 56 |
property. This might look something like this:: |
| 57 |
|
| 58 |
class Image(HasTraits): |
| 59 |
... |
| 60 |
|
| 61 |
histogram = Property(Array, depends_on='image') |
| 62 |
|
| 63 |
def _get_histogram(self): |
| 64 |
hist, bins = np.histogram( |
| 65 |
self.image, |
| 66 |
bins=256, |
| 67 |
range=(0, 256), |
| 68 |
density=True, |
| 69 |
) |
| 70 |
return hist |
| 71 |
|
| 72 |
This works, but it has a downside that the histogram is re-computed every |
| 73 |
time the property is accessed. For small images, this is probably OK, but |
| 74 |
for larger images, or if you are working with many images at once, this may |
| 75 |
impact performance. In these cases, you can specify that the property |
| 76 |
should **cache** the returned value and use that value until the trait(s) |
| 77 |
the property depend on changes:: |
| 78 |
|
| 79 |
class Image(HasTraits): |
| 80 |
... |
| 81 |
|
| 82 |
histogram = Property(Array, depends_on='image') |
| 83 |
|
| 84 |
@cached_property |
| 85 |
def _get_histogram(self): |
| 86 |
hist, bins = np.histogram( |
| 87 |
self.image, |
| 88 |
bins=256, |
| 89 |
range=(0, 256), |
| 90 |
density=True, |
| 91 |
) |
| 92 |
return hist |
| 93 |
|
| 94 |
This has the trade-off that the result of the computation is being stored |
| 95 |
in memory, but in this case the memory is only a few hundred bytes, and so |
| 96 |
is unlikely to cause problems; but you probably wouldn't want to do this |
| 97 |
with a multi-gigabyte array. |
| 98 |
|
| 99 |
Exercise |
| 100 |
-------- |
| 101 |
|
| 102 |
Make ``pixel_area`` a read-only property. |
| 103 |
|
| 104 |
""" |
| 105 |
|
| 106 |
import os |
| 107 |
import datetime |
| 108 |
|
| 109 |
from PIL import Image as PILImage |
| 110 |
import numpy as np |
| 111 |
|
| 112 |
from traits.api import ( |
| 113 |
Array, Date, File, Float, HasTraits, Property, Str, Tuple, |
| 114 |
cached_property, observe |
| 115 |
) |
| 116 |
|
| 117 |
|
| 118 |
class Image(HasTraits): |
| 119 |
""" An SEM image stored in a file. """ |
| 120 |
|
| 121 |
filename = File(exists=True) |
| 122 |
sample_id = Str() |
| 123 |
operator = Str("N/A") |
| 124 |
date_acquired = Date() |
| 125 |
scan_size = Tuple(Float, Float) |
| 126 |
|
| 127 |
scan_width = Property(Float, depends_on='scan_size') |
| 128 |
scan_height = Property(Float, depends_on='scan_size') |
| 129 |
|
| 130 |
image = Array(shape=(None, None), dtype='uint8') |
| 131 |
histogram = Property(Array, depends_on='image') |
| 132 |
|
| 133 |
pixel_area = Float() |
| 134 |
|
| 135 |
# Trait observers |
| 136 |
|
| 137 |
@observe('filename') |
| 138 |
def read_image(self, event): |
| 139 |
pil_image = PILImage.open(self.filename).convert("L") |
| 140 |
self.image = np.array(pil_image) |
| 141 |
|
| 142 |
# compute some extra secondary attributes from the image |
| 143 |
if self.image.size > 0: |
| 144 |
self.pixel_area = ( |
| 145 |
self.scan_height * self.scan_width / self.image.size |
| 146 |
) |
| 147 |
else: |
| 148 |
self.pixel_area = 0 |
| 149 |
|
| 150 |
# Trait default methods |
| 151 |
|
| 152 |
def _date_acquired_default(self): |
| 153 |
return datetime.date.today() |
| 154 |
|
| 155 |
def _scan_size_default(self): |
| 156 |
return (1e-5, 1e-5) |
| 157 |
|
| 158 |
# Trait property methods |
| 159 |
|
| 160 |
def _get_scan_width(self): |
| 161 |
return self.scan_size[0] |
| 162 |
|
| 163 |
def _set_scan_width(self, value): |
| 164 |
self.scan_size = (value, self.scan_size[1]) |
| 165 |
|
| 166 |
def _get_scan_height(self): |
| 167 |
return self.scan_size[1] |
| 168 |
|
| 169 |
def _set_scan_height(self, value): |
| 170 |
self.scan_size = (self.scan_size[0], value) |
| 171 |
|
| 172 |
@cached_property |
| 173 |
def _get_histogram(self): |
| 174 |
hist, bins = np.histogram( |
| 175 |
self.image, |
| 176 |
bins=256, |
| 177 |
range=(0, 256), |
| 178 |
density=True, |
| 179 |
) |
| 180 |
return hist |
| 181 |
|
| 182 |
|
| 183 |
# --------------------------------------------------------------------------- |
| 184 |
# Demo code |
| 185 |
# --------------------------------------------------------------------------- |
| 186 |
|
| 187 |
this_dir = os.path.dirname(__file__) |
| 188 |
image_dir = os.path.join(this_dir, "images") |
| 189 |
filename = os.path.join(image_dir, "sample_0001.png") |
| 190 |
|
| 191 |
# load the image |
| 192 |
image = Image( |
| 193 |
filename=filename, |
| 194 |
operator="Hannes", |
| 195 |
sample_id="0001", |
| 196 |
) |
| 197 |
|
| 198 |
# perform some sample computations |
| 199 |
print("Scan size:", image.scan_size) |
| 200 |
|
| 201 |
print("Change scan width to 1e-4") |
| 202 |
image.scan_width = 1e-4 |
| 203 |
|
| 204 |
print("Scan size:", image.scan_size) |
| 205 |
|
| 206 |
for filename in os.listdir(image_dir): |
| 207 |
if os.path.splitext(filename)[1] == '.png': |
| 208 |
print() |
| 209 |
print("Changing filename to {}".format(filename)) |
| 210 |
image.filename = os.path.join(image_dir, filename) |
| 211 |
|
| 212 |
print( |
| 213 |
"The most frequent intensity of {} is {}".format( |
| 214 |
filename, |
| 215 |
image.histogram.argmax(), |
| 216 |
) |
| 217 |
) |